[Radiance-general] Shadow above highlight?

[Greg Ward]

Hah -- looks like Andy arrived at the same conclusion, though a bit  
more speculatively.  Well done!

There are methods for sampling a hemisphere without the singularity at  
the zenith, but they are a lot trickier to implement.

-Greg

> From: Andrew McNeil <andrew.mcneil at arup.com>
> Date: August 13, 2009 4:31:25 PM PDT
>
> John,
>
> I'm running your scene and am witnessing the same effect.   
> Interestingly the
> dark spot still exists, it is just much less pronounced.
>
> I am both able and unable to understand why this works.
>
> Years ago I remember someone mentioning the impossibility of ambient
> sampling normal to a surface and it's lingered in my mind ever  
> since.  I've
> always struggled with the ambient calculation's ability to sample the
> horizon which technically contributes nothing to illuminance at a  
> point
> while it was unable to sample the zenith which contributes most to
> illuminance.
>
> I've thought about suggesting the range for ambient sampling switch  
> from
> [0,1) to (0,1]  but couldn't grasp the ramifications of the  
> possibility of
> sampling the zenith from all of the zenith adjacent ambient divisions.
>
> It seems to me that statistically it isn't possible for this to make a
> difference so maybe there is a bug in the ambient super sampling,   
> but it
> also seems to me that statistically this must make a difference.
>
> So I guess I'll just have to wait for Greg and others to weigh in...
>
> Andy